645 university-level Poznań students of English answered a questionnaire concerning their perception of English pronunciation, as taught in their respective schools (School of English, School of Russian, Teacher-Training College). Most (a) have used some phonetic transcription in their notes and believe that it is not too difficult for learners, (b) wish they had more pronunciation practice, (c) believe that listening to English is not more difficult than speaking it and that good English pronunciation is not more important than grammar or vocabulary. About half of the respondents have no opinion on whether American respelling transcriptions are easier than British, and they are divided equally on whether or not they practise English pronunciation regularly on their own.

Two independent variables: the respondent's respective school and the year of study s/he is in were chosen for closer scrutiny. As far as the former is concerned, the statistically significant differences in frequency distributions observed are clearly due to (a) the socio-economic background of the students and (b) the character of the school. The year of study was found to affect especially (a) active use of phonetic transcription and pronunciation practice outside of class (inversely) and (b) assessment of transcription difficulty (directly).

Some conclusions are drawn, on a macro (educators should take it into account that Polish students believe English pronunciation to be important, and are not afraid of phonetic transcription) as well as on a micro scale (Russian School English teachers can benefit from their students' very good attitude towards transcription).

Sixty-nine first year students of the School of English, Adam Mickiewicz University, filled in a "computer file extension questionnaire" containing twelve monomorphemic English words with medial -C₁C₂C₃- clusters. They were asked to "choose one of the provided extensions for a given word" to "help computer scientists to create good, user-friendly extensions".

The real aim of the questionnaire was to see which bisegmental cluster they would rather choose, in constructing a CCC extension, to complement the word-initial C₀: -C₁C₂- or -C₂C₃-. It was hypothesised that their choice might be at least partly controlled by the phonotactics of the clusters, be it language-specific phonotactics of English or the universal phonotactics, based on optimal sonority relations between segments.

The overriding principle appears to be to take the leftmost pair, -C₁C₂-. Extension coining is thus shown to follow certain general psychophonetic and metaphonological regularities, such as word-onset salience and mnemonicity. Simultaneously, however, (1) the selected -C₁C₂- clusters are all phonotactically correct word-final clusters in English, (2) most of the selected -C₂C₃- clusters are phonotactically correct word-initial clusters in English, (3) while the coined "triples" are all universally preferred word-medial clusters (not all of them attested in English).

Lexicographic phonetics is phonetics applied to the process of dictionary-making. While it has not been labeled as such, it has traditionally been concerned with issues such as: the choice of accent and transcription to represent in dictionaries, the extent of dialectal, phonostylistic and idiosyncratic variation of pronunciation covered, the representation of stress and weak forms, etc. Authors have included Abercrombie, Gimson and Wells, among others.

Phonetic lexicography is what dictionary makers and critics do when they ponder sound representation from the lexicographic perspective. Issues of relevance include: the questions of consistency, the place and role of pronunciation in the microstructure of the dictionary, the treatment of pronunciation in learners' dictionaries, sound recording, playback and synthesis in electronic multimedia dictionaries, and others. Few (meta)lexicographers have ever done substantial work in this area.

Ultimately, the two pursuits cannot be clearly delimited, of course. They both cover, each from its own perspective, the little-explored ground of sound representation in dictionaries. Both the traditionally tackled issues and the new vistas are discussed in the paper.

A new pronunciation standard has recently been proposed as a viable goal for foreign learners of English to reach, that of Lingua Franca Core (Jenkins 2000). LFC would be the pronunciation of English as an International Language (EIL) or English as a Lingua Franca (ELF). The project for a 'realistic' pronunciation syllabus for International English has gained wide support in many EFL circles. Yet, it is marred by a number of faults and weaknesses. In this presentation arguments are provided against LFC-like standards/syllabuses, and for the traditional RP/GA ones. The arguments come from a wide range of considerations: (1) philosophical: axiology does not follow from ontology, (2) ideological: political correctness (anti-linguistic-imperialism and the like) does not serve objective scientific judgement, (3) (socio)linguistic: speech is used for much more than sheer communication, (4) pedagogical: loosening didactic standards must not be post-hoc excused and encouraged, (5) psychological: native-like pronunciation is recognized by learners as an asset in itself. They are developed in reaction to the writings and pronouncements of the main proponents of LFC pronunciation: Jennifer Jenkins, Barbara Seidlhofer, and others. All argumentation is relativized to the here-and-now of the Polish EFL scene at the onset of the 21st century.

Jenkins,J. 2000. The phonology of English as an international language. Oxford: OUP.

Modern Machine-Readable Dictionaries (MRDs) offer users an unprecedented richness of content and form, and gradually oust traditional paper-based word books out of existence. Despite the breathtaking developments of hard- and software, however, popular MRDs, especially those made for learners of foreign languages, are still deficient in a number of respects. Two of these are dealt with in this paper: (a) width and flexibility of user access to the riches of lexicographic content, and (b) the degree and (artificial) 'intelligence' of user modelling and customisation. It is argued that the two deficiencies are not due to any inherent technological obstacles, but rather to the conservatism of dictionary makers and users (both learners and teachers). A few examples of functionalities 'which could be, but are not' are provided in a hypothetical case study of EFL student Tom and his MRD.

Macmillan English Dictionary on CD-ROM is one of few monolingual EFL electronic dictionaries on the market to attempt to seriously address the phonetic potential of the lexicographic content and the multimedia technology. This is a detailed analysis of the structure and functioning of the phonetic representation in CD-ROM MED, with postulates and conclusions generalized to all learner's machine-readable dictionaries (MRDs). The three essential elements of MED phonetics are discussed: (a) the phonetic representations, both transcription and audio recordings, (b) the phonetic access mechanism through SoundSearch, (c) the Pronunciation Practice module. Some phono-lexicographic problems with the three components are pointed out.

Postcards, advertising leaflets and billboards, xerox-, SMS- and internet-lore can be ample sources of linguistic trivia of all kinds, including those where some aspects of pronunciation are brought to the fore, usually through punning. These materials can, and should, be used to boost the metaphonetic competence (awareness, intuition) of foreign language learners in all areas of segmental and suprasegmental phonetics. This metacompetence is instrumental in teaching and learning foreign pronunciation, as part of the declarative-knowledge component thereof. Examples are provided of many categories of metaphonetic trivia.

State of the art in pronunciation-oriented EFL CALL is reviewed from the pedagogical perspective. Discussion touches upon CALL flexibility, coverage, declarative vs. procedural knowledge, L1-sensitivity, multimedia employment and automatic speech recognition (ASR). Six different CALL programs are briefly evaluated from these points of view: Fluency, Pronunciation Power, Connected Speech, Better Accent, ISLE and Tell Me More. Future promises and challenges in speech-enabled EFL CALL are outlined, such as speech synthesis, multimodality in man-machine communication and (speech-to-speech) machine translation.

Rule-based Text-to-Speech synthesis (TTS) is discussed from the point of view of English as a Foreign Language (EFL) Computer-Assisted Language Learning (CALL). The perspective is pedagogical rather than technological. Some didactically salient characteristics of TTS are considered, such as (a) its feasibility as a pronunciation model, (b) control afforded over accentual and phonostylistic variation of speech, (c) the prospects of multimodal synthesis ('talking heads'). Some internet website addresses featuring TTS information, products and demos are provided.

In my Euralex 2000 Stuttgart contribution I analysed phonetic keywords found in the pronunciation guides to 'the big four' monolingual learners' dictionaries of English (OALDCE4, LDOCE3, COBUILD1, CIDE). These keywords, whose function it is to cue unfamiliar IPA symbols used in the dictionaries' phonetic representations, were evaluated with respect to their native Paivio-familiarity, EFL phonetic difficulty and textual frequency. It was shown that while they were more or less equivalent on the first two scores, dictionaries varied widely in terms of the third. Appeal was made for more research into the phonetic structure and choice of keywords.

With the recent 2002 appearance of the fifth 'big', the Macmillan English Dictionary for Advanced Learners (MED), I used the previously applied yardsticks to measure how the new kid on the block fares. The answer is: (a) just like the others in terms of keyword familiarity, (b) worse in terms of phonetic difficulty, and (c) above average in terms of frequency. Some unique features of MED's keywords vis-a-vis the other four EFL dictionaries are pointed out.

A phonostatistic analysis of pronunciation difficulty in the 88495 definitions of the British edition of Macmillan English dictionary for advanced learners (MEDAL) is performed. The L1-sensitive (Polglish) Phonetic Difficulty Index (PDI) of each word in every definition is automatically calculated, summed and averaged, as is the definition word-length (mean=11.91). Selected definitions are compared across three dictionaries: MEDAL, CALD and LDOCE4. Some defining vocabulary (DV) phonostatistics are also provided for the three dictionaries (PDI frequency distributions and means). Appeal is made for more control over definition phonetics, which is claimed to be important, especially for beginning and intermediate learners, who tend to subvocalize in reading.

The Phonetic Difficulty Index (PDI) is discussed, its origin, bibliography, structural design, functionalities and present as well as potential applications. PDI is a global numerical measure of lexico-phonetic difficulty of English words to Polish EFL (Polglish) learners. It originated as part of the electronic Phonetic Access Dictionary (PAD) in the mid-1990's. It accounts for learners' common (a) grapho-phonemic, (b) L1-transfer, and (c) L2-interference pronunciation problems. PDI is currently programmed as a computer algorithm working on alphanumeric ASCII strings. It can be used for querying electronic dictionaries, in semi-automatic pronunciation exercise generation, in compilation of word-lists for EFL instruction and psycholinguistic experimentation, for NLP applications, as well as in phonetic and phonological research.

Corpora of English text (both native and non-native) are now taken for granted as a resource in teaching and learning English as a Foreign Language (EFL). So far they have, however, been exploited mostly on the lexical, morpho-syntactic and stylistic level. The phonetic potential of raw-text corpora (as opposed to the few expensive acoustically treated and annotated ones) has not been discovered.

In this contribution a method is presented of automatically annotating raw-text corpora with EFL phonetic tags coming from a suitably treated electronic word-list. The focus of the presentation is in phonetic lapsology, i.e. in annotating English text for probable Polglish (Polish-English interlanguage) pronunciation problems and errors, as well as for the overall level of pronouncing difficulty. Two examples from my research are presented and discussed: (1) phono-lapsological analysis of definitions in Macmillan English Dictionary for Advanced Learners on CD-ROM (MEDAL; Sobkowiak, forthcoming) and (2) my current work on TIMIT sentences in the context of the Boulder-Poznań CSLR Colorado Literacy Tutor project. It is demonstrated that on top of automatic phonetic transcription of raw text, which is now conceptually and technologically rather trivial, a sophisticated L1-sensitive automatic phonetic annotation is feasible, with a variety of EFL-related functions, in particular text/sentence selection (e.g. TIMIT) and evaluation (e.g. MEDAL) for lexicographic, pedagogical and research (Sobkowiak, unpublished) purposes.

The query facilities built into electronic dictionaries of English as a foreign language (EFL) have significantly improved in recent years: hypertextual links, global text searches, Boolean search criteria combinations, access through phonetic transcription, etc. However, as far as dictionary content representation is concerned, the graphical user interfaces (GUI) mostly keep to the old tradition of two windows: (1) dynamically searchable macrostructural wordlist on the left, and (2) static microstructural entry panel on the right. Other boxes, menus, frames and flashcards are just embellishments of this fundamental standard interface scheme. Pronunciation in particular is graphically represented in but one century-old method: as a phonetic transcription field appearing (or not) right after the entry headword. In this paper I argue for the application to the representation of EFL electronic dictionary pronunciation of the new Java- and Flash-enabled animated GUIs, recently proposed for database querying by various authors. The benefits to learners include: (a) better mnemonicity, (b) improved customisability, (c) direct intuitive searchability, (d) explicit mapping of phonetic processes, relations and groupings (homophony, allophony, similarity, minimal-pairs, assimilations, deletions, L1 substitutions, etc.), and others.

Postcards, advertising leaflets and billboards, xerox-, SMS- and internet-lore can be ample sources of linguistic trivia of all kinds, including those where some aspects of style and dialect are brought to the fore, usually through punning. These materials can, and should, be used to boost the stylistic awareness of foreign language learners on all levels of language structure and function: from phonetics to pragmatics. This awareness, or stylistic metacompetence, is instrumental in teaching and learning foreign language for communication, as part of the declarative-knowledge ("know-that") component thereof. Examples are provided of many categories of such stylistic trivia.

Monolingual foreign language dictionaries for learners are full of definitions. Normally, these are used, both by learners and teachers, for one purpose only: ascertaining the meaning of the currently looked-up lexical item. In electronic dictionaries one would also normally be allowed to run the so-called 'definition search', whereby headwords having a particular string in their definition would be listed for further inspection. But even in the most sophisticated electronic dictionaries for FL learners definitions (as well as example sentences) remain almost completely unexploited as a corpus of FL text.

In this paper, definitions taken from MEDAL (British English version) are scrutinized for the EFL potential which they hold across a variety of linguistic levels: from pronunciation, through vocabulary and syntax, to style. Treated as a properly tagged corpus of English text (about one million words in all), MEDAL definitions can be forged into an extremely rich reservoir of EFL insights, tasks, tests and exercises.

Sobkowiak shows how phonetically transcribed and pronouncing-difficulty-annotated definitions can be used to focus on selected EFL phonetic problems, with a built-in measure of L1-sensitivity. Kaszubski argues that POS-tagged dictionary definitions can help students acquire difficult lexico-stylistic distinctions and avoid the notorious (but hard-to-pinpoint) under- and over-use of some words and collocations. Wojnowska implements these insights in an automatic corpus-based TestBuilder of her own design, a program which assists teachers and learners in test preparation in a variety of formats (hangman, gaps, etc.).

It is shown that, properly annotated and exploited, electronic dictionary definitions need not function for reference only, but can significantly boost the didactic value of the dictionary. Such a teaching resource built into learners' dictionaries would follow the current trends in pedagogical lexicography, and foreign language teaching generally, where the lexicon is seen as the hub of FL skills, and the dictionary gradually takes the central position among all FL resources.

A lexical database of English originating from the machine-readable version of OALDCE has been annotated with two types of Polglish pronouncing difficulty tags: (a) aggregate numerical measures of grapho-phono-lexical difficulty, the Phonetic Difficulty Index (PDI), with a range of 0-10, and (b) qualitative PDI codes of specific grapho-phonetic difficulty, selected from a list of 57 such items, derived empirically from teaching experience. Some PDI codes (i.e. some grapho-phonetic difficulties) tend to cooccur in English words with higher than chance frequency. Words identified in this manner as 'knots' of the difficulty network can then be used for EFL teaching, dictionary making, natural language processing, linguistic experimentation and theory building, etc. Further applications of PDI-code cooccurrence are foreseen, such as building a lattice of qualitative phono-lapsological lexical equivalence within the English lexicon.

1. The defining vocabularies of the EFL dictionaries do not differ significantly among themselves in phonolapsological terms. None is on average phonetically harder than the reference lexicon.
2. The phonostatistic profile of MEDAL definitions on the background of reference phonolexical data of English, in terms of phoneme frequency, shows that the differences are mostly due to the incidence of specific DV words: those very heavily used in definitions (for their grammatical or definitional value) boost the frequency of phonemes contained in them. Some of these phonemes are inherently phonetically difficult.
3. The word-weighted mean phonetic difficulty of MEDAL definitions is significantly lower than reference, with local maxima due to the choice of specific DV items, as above, or to the shortness of some definitions which function as cross-reference anchors rather than definitions proper.
4. A fair proportion of global phonetic difficulty encountered in MEDAL definitions is due not to phonolexical problems as such, but rather to a host of prosodic and sandhi phenomena (intonation, stress, rhythm, assimilation, reduction, deletion) arising when words are used in context. Only some of these are captured by the (current version of) PDI metric.
5. A PDI analysis conducted across the sample of definitions coming from the five EFL dictionaries shows that, while there are serious discrepancies in the handling of specific definitions, the average PDI figures are strikingly similar across the board.
6. A micro-analysis of a number of definitions from the point of view of their phonetic difficulty to EFL learners shows that there are systematic ways in which dictionary makers can control definition phonetics more tightly than has so far been the case, for example by phonetically sensitive DV word selection or by paying more attention to juncture phenomena.

Włodzimierz Sobkowiak's full bibliography